The pursuit to creating the most human-like AI

From the rise of the virtual influencer to the launch of Samsung’s NEON at CES this year, 2020 has seen an increasing shift towards humanising Artificial Intelligence (AI).

The technology is becoming so lifelike, you could mistake some robots for real people.

With Forrester predicting that one-tenth of start-ups will emerge with more digital workers than human ones, can we expect these artificial beings to become lifelike colleagues in the not too distant future? Can this new breed of AI pass the Turing test? And how can we distinguish between the AI assistants that will deliver tangible value to our lives and work and those that are nothing more than a fun gimmick?

A vision of the future

Pop culture has tricked us into thinking that human-like AI is an impending threat to humankind. How many films or TV programmes have you seen where AI and robotics combine to create artificial beings that could be mistaken for humans and then, inevitably, turn catastrophically destructive? Despite the cultural misrepresentation, the tech industry is in fact moving towards a future whereby AI assistants that look more and more like us.

But it’s not about being just visually appealing. For humans to really trust AI assistants they need to have emotional and cognitive capabilities too. For example, positive nonverbal cues, such as maintaining eye contact and being expressive with face and hand gestures, help us to connect and engage in conversation. These are the small, oft forgotten, features that companies developing AI should strive for in order to emulate real, lifelike interaction and invoke trust. 

The power of cognition 

The vast majority of us already interact with AI assistants daily, whether through a smart home device or a customer service chatbot. Most of us have also experienced the frustration that digital assistants can sometimes incite by misunderstanding queries and carrying out the wrong requests. If an AI assistant has no cognitive power, then we’re far from passing the Turing test as we need to match brains with beauty. As such, we need to look beyond just a pretty face to determine the future of general intelligence. Hyper-realistic avatars with no cognitive capabilities, such as emotional intelligence and context switching, are as valuable as the limited, static chatbots that cause so many of us to leave customer service queues.

Defining true artificial intelligence

What customers want, and society needs, is an AI-powered assistant capable of having a human-like conversation and understanding intent – no matter how you phrase it. By gathering deep contextual understanding, a cognitive AI assistant can easily understand intent and quickly navigate toward a solution. A system displays true intelligence by understanding the concepts behind the actions it is taking. 

For example, let’s say you have £300 in your current account, and you transfer £200 into your savings account. How much money do you have left? An AI assistant won’t come to the answer of £100 using pattern recognition; rather, it understands the underlying concept of what “transfer” means and then applies the concepts of subtraction. 

Applying the rules to the workplace

In a workplace setting, an AI assistant’s ability to understand and respond to a query or request quickly and appropriately is the utmost important skill. Currently, cognitive AI systems are able to handle thousands of customer service queries, at scale, across multiple languages. They can understand context switching, are able to field multiple queries at once and escalate when appropriate – the most advanced AIs are already processing complex requests like mortgage applications for large banks.

So, why do they need to look good too? These use cases can be further enhanced with AI assistants that have a human-like interface that can build an empathetic connection with the customer. If a stressed or angry customer feels like their complaint or query has been fully understood and registered, they’re more likely to look favourably on that company in future. There’s no point in having an AI that appears lifelike unless they also have the brainpower to respond intelligently.

Making the hybrid workforce vision a reality

The future is likely to include AI assistants who not only look like us, but also sound and act like us and interact in very human ways. We are likely to see the emergence of hybrid teams made up of both human and digital employees, acting in roles as varied as L&D Coordinator and Mortgage Advisor. 

Human-like AI will be one of the most transformative technologies of the next ten years, likely making sci-fi films a reality. Yet, we need to ensure that the AI we create has the in-built emotional intelligence to conduct a real, empathetic and human-like conversation. This is the bar we should be holding digital beings to if we want them to have meaningful interactions with humans and not just be a pretty, digital face.

About the Author

Faisal Abbasi is Managing Director Western Europe & MEMA, Amelia (an IPsoft company).  Digital Colleagues and virtual engineers will revolutionize human experiences through the rapid delivery of personalized services. No company is more prepared to lead this march toward a collaborative future than Amelia, an IPsoft Company.